Upwork's (UPWK) CEO Stephane Kasriel on Q3 2018 Results - Earnings Call Transcript

|
About: Upwork (UPWK)
by: SA Transcripts

Upwork, Inc. (NASDAQ:UPWK) Q3 2018 Earnings Conference Call November 7, 2018 7:00 AM ET

Executives

Palmira Gerlach – Director-Investor Relations

Stephane Kasriel – Chief Executive Officer

Brian Kinion – Chief Financial Officer

Analysts

Mark May – Citi

Mark Mahaney – RBC Capital Markets

Brent Thill – Jefferies

Ron Josey – JMP Securities

Scott Devitt – Stifel

Nandan Amladi – Guggenheim Partners

Operator

Good day, ladies and gentlemen, and welcome to the Upwork Q3 2018 Earnings Conference Call. At this time, all participants are in a listen-only mode. Later we will conduct a question-and-answer session and instructions will follow at that time. [Operator Instructions] As a reminder, this call may be recorded.

I would now like to introduce your host for today’s conference, Palmira Gerlach, Director of Investor Relations. You may begin.

Palmira Gerlach

Hello and welcome to Upwork’s discussion of its third-quarter 2018 financial results. Leading the discussion today are Stephane Kasriel, Upwork’s Chief Executive Officer, and Brian Kinion, Upwork’s Chief Financial Officer. Following management’s prepared remarks, we will be happy to take your questions.

But first, let me review the Safe Harbor statement. During this call, we may make statements related to our business that are forward-looking statements under the federal securities laws. These statements are not guarantees of future performance, but rather are subject to a variety of risks, uncertainty, and assumptions.

Our actual results could differ materially from expectations reflected in any forward-looking statement. For a discussion of the material risks and other important factors that could affect our actual results, please refer to our SEC filings available on the SEC’s website and on our website as well as the risks and other important factors discussed on today’s earnings release.

In addition, reference will be made to non-GAAP financial measures. Information regarding reconciliation of non-GAAP to GAAP measures can be found on the press release that will be issued this afternoon on our investor relations website.

Please note that the prepared remarks corresponding to the information reviewed on today’s conference call will also be available on our investor relations website at investors.upwork.com once the call has concluded.

With that, I will turn the call over Stephane.

Stephane Kasriel

Good afternoon and thank you, everyone, for joining us for our first quarterly earnings call as a public company. We are excited to share our progress with you and to provide details on our financial performance.

But before I do, I would like to take a moment to express my gratitude to our investors, our clients, the freelancers on our platform, the freelancers who provide services to us, and our employees. Each of you has put your passion and trust in us over the years and I greatly appreciate it.

We were pleased with our IPO process and investors’ reception to our offering. We are excited to share today that we exceeded our own expectations for the third quarter of 2018. While our IPO was a major milestone for Upwork, it was just another step in the evolution of the Company. We are addressing a large and expanding market opportunity, of which we believe we have merely scratched the surface.

Since many of you are new to the Upwork story, I wanted to provide a brief review of our business and our opportunity before we look more closely at the progress we made in the third quarter. Upwork’s mission is to create economic opportunities so people have better lives.

The world is in the midst of a transformation in how we work. Technology is enabling people to work how, where, when, and with whom they please. Upwork is at the forefront of addressing this opportunity as our platform removes the constraint of location, making hiring faster and more efficient.

We operate the largest online marketplace for highly skilled freelance talent, as measured by gross services volume, or GSV. As a reminder, GSV is the total amount of business transacted through our platform. As we last disclosed, over 475,000 clients hired more than 375,000 freelancers in 180 countries through our platform in the last 12 months.

Now that you have heard about the Upwork story, let’s talk about what is going on in the freelance market. First, when you look at the number of service jobs that can be performed remotely, we believe we are addressing a large and expanding market opportunity, which we estimated to be USD560 billion of GSV in 2017.

Upwork’s success would not be possible without the highly skilled freelancers that use our platform routinely. For these freelancers, Upwork offers a full value proposition: the flexibility to choose their clients and projects, to negotiate their own rates, and to manage their personal schedules on their own terms.

We offer security and timeliness of payments, provide strategic insight into data relevant to the individual freelancer, and help the freelancer build their brand and reputation. Upwork serves as a powerful marketing channel for freelancers to advertise their businesses.

Exactly a week ago, the fifth annual Freelancing in America survey, or FIA, was released. The FIA survey is the most comprehensive measure of the U.S. freelance workforce. It is conducted by independent research firm Edelman Intelligence, surveying more than 6,000 U.S. workers.

We cosponsor the survey along with Freelancers Union, the largest U.S. organization representing independent workers. A five-year scorecard of the U.S. freelance workforce includes these findings, which bode well for the market as well as the customers we serve.

For one, the market is growing quickly and evolving from traditional work. In the last five years, the growth rate of the U.S. freelance workforce was over three times greater than the growth rate of the traditional U.S. workforce. The study found that one in three American workers or approximately 56 million people freelance.

Also, freelancers are dedicating more of their bandwidth to freelancing. 28% of freelancers are now full-time versus only 17% back in 2014. In addition, freelancing is being driven by Gen Z and millennials. These generations are freelancing more than any prior generation, with 45% of Gen Z and 42% all of millennials freelancing. And finally, technology is a powerful tailwind. This year, 64% of freelancers had obtained a project online versus only 42% five years ago, a leap of 22 points.

As technology helps freelancers connect to do more work, it’s also helping businesses find freelancers with cutting-edge skills. According to Upwork’s quarterly skills index, the fastest-growing skills on Upwork include many innovative technological skills, such as blockchain, augmented reality, and machine learning. Fast-growing non-tech skills include social media management, brand strategy, and influencer marketing.

On the client side, the value proposition stems from the access to top talent, reduced time to hire, cost savings, and the efficiencies of streamlined workflows. Clients often receive multiple bids within minutes of posting a job with the median time to hire being 23 hours in 2017.

Our deep data set, search, and machine learning algorithms are the backbone of the speed in matching. Additionally, the Upwork platform software streamlines the interviewing, streaming, and contracting process. Our marketplace’s direct-to-talent approach reduces the need for costly and inefficient intermediaries such as staffing firms, recruiters, and traditional agencies.

We continue to see strong adoption on the client side. Growing existing small business clients spend on the platform and acquiring new small business clients remains a priority for us, as they represented approximately 80% of our GSV in 2017 and are the vast majority of our customers.

We continue to make progress in growing our direct sales efforts, which are focused on mid-market and large business clients with 100 employees or more. As of Q3, we had more than 30% of the Fortune 500 using Upwork products, whether through our Upwork Standard or our Upwork Enterprise offering.

The priority for our sales team is to grow Upwork Enterprise clients through a land-and-expand strategy. In the last 12 months, we have had many customers that have spent millions of dollars on our platform across consumer packaged goods, technology, and media verticals and we are just getting going in the Enterprise space.

We are continuing to bring these large businesses together to have high-level discussions about the future of work and help guide them on how to change their talent strategies. In August, we hosted our second-annual Work Without Limits Executive Summit, which brought together over 100 executive leaders from Fortune 500 companies to discuss how they are embracing flexible teams.

Now I’d like to discuss some of our product and engineering accomplishments that will drive our larger business goals. In this vein, we added features to further enhance viral loops within our marketplace product offerings. One is called Help Me Hire and is targeted at encouraging clients to invite additional hiring managers within their organization to review a potential hire.

We also soft launched the My Jobs Dashboard. This new feature will make it easier for clients to get inspiration for new projects and manage current projects by sharing job posts and favorite freelancers. Both of these features are examples of our expand existing client strategy through product innovation.

Our most successful enterprise clients run a flexible talent program within their organizations. In Q3, we launched a program owner experience within the Upwork Enterprise product. This new feature provides larger companies with the resources and insights they need to build, scale, and measure the impacts of their flexible talent program.

While we’ve invested in mobile for years, our approach had been to fund our mobile web, iOS, and Android applications as separate teams so that we could build the best experience. It was difficult to ensure that all of our product functionality was available across devices.

In 2018, we spent much of our engineering efforts year to date transforming our product to be mobile-first. We have migrated 90-plus of our most important web pages to be accessible via mobile browsers to ensure product parity across all endpoints. This will be an ongoing effort that affects how we make decisions about all elements of our product roadmap moving forward.

We also plan to focus on and verticalize the more than 5,000 skills offered across the 70 categories on our platform, as the ability for clients to hire across multiple categories is a key differentiator for us. We will also continue to invest to further enhance the software and functionality of the platform, as we are always trying to improve the user experience on our website and mobile apps.

Additionally, we continue to invest in trust and safety capabilities to stay a step ahead in making our platform the most trusted to get work done. In short, we are just getting started in changing the world of work and that is what drives us at Upwork.

Before I turn the call over to Brian, I wanted to do once again thank everyone who made our IPO a reality. From the early days of oDesk and Elance to the united front of Upwork, many years of tireless innovation and collaboration from our dedicated employees and Upwork’s internal team of freelancers made our IPO last month a possibility.

Also, to our clients, to the freelancers who find work on our platform every day, and to our investors, we truly appreciate the confidence you have placed in Upwork and look forward to achieving great milestones together in the years to come. We thank you.

And with that, I would like to hand it over to our Chief Financial Officer, Brian Kinion.

Brian Kinion

Thank you, Stephane. Good afternoon, everyone. My remarks today will start with a brief update on our key operating metrics for the third quarter, then turn to the financial results and our guidance for the fourth quarter that we provided in our earnings release filed today.

Numbers are rounded for the sake of convenience, and I will make comparisons on a year-over-year basis unless otherwise noted. I’ll be referring to GAAP measures unless explicitly cited as a non-GAAP measure.

We monitor and measure our business performance using certain key operating metrics, which include GSV, core clients, and client spend retention. We believe these metrics are key indicators of our growth and the overall health of our business.

GSV, which includes both client spend and additional fees that we charge for other value-added services, increased by 27% in Q3 to $449 million. The growth in GSV was driven by an increase in core clients and an increase in our client spend retention.

Growth in core clients is an important metric for two reasons. Core clients have been historically more likely to continue using our platform; and two, they represent approximately 80% of our GSV. We define a core client as a client that has spent in aggregate at least $5,000 in their lifetime on our platform and have spent in the last 12 months.

As of September 30, 2018, we had approximately 101,000 core clients, representing a 22% increase. Client spend retention was 108% on a trailing 12-month basis at September 30, 2018, compared to 95% at September 30, 2017.

Client spend retention is derived by dividing recurring client spend by our base client spend from the same clients over the four quarters ended one year from the date of measurement. We are pleased that client spend retention has continued to improve over time and has reached its highest level since the combination of Elance and oDesk in 2014.

Client spend retention illustrates the recurring nature of our business, even though clients are not contractually required to spend on a recurring basis. We expect client spend retention to stabilize in the 106% to 108% range for the near term based upon our analysis of our current cohorts. However, we are focused on increasing recurring spend from existing clients on our platform by building new products, features, and functionality as well as marketing and sales efforts.

Now let me provide some additional insights into our business model. We monetize by charging both freelancers and clients different fees in our two-sided B2B marketplace. Our overall take rate, which we define as total revenue as a percentage of GSV, has multiple components and can fluctuate from period to period, impacting revenue growth rates as well as gross margins.

Let me explain the three largest drivers of our take rate today in more detail. First, the majority of our total revenue is generated from our marketplace offerings, where we recognize revenue on a net basis. We also have a managed services offering where we engage freelancers for projects, invoice the client directly, and assume responsibility for the work performed, from which we are required to recognize the revenue on a gross basis. Therefore, GSV and revenue are the same amount for our managed services offering.

The mix of revenue generated in a quarter from our marketplace offerings versus our managed services offering will impact overall take rate. Our marketplace revenue grew at a faster rate year over year in the third quarter as compared to our managed services revenue, which resulted in a lower overall take rate, but generated a higher gross margin.

Second, for our Upwork Standard offering, we charge freelancers a tiered service fee, which we introduced in June of 2016. We charge 20% for the first $500 for each unique freelancer/client relationship; 10% for the next $9,500; and then 5% thereafter.

We have seen over time more of these unique relationships getting to the 5% billing tier. We view this as a positive trend for our business, as the goal of the pricing change was to incent long-term relationships and recurring use of our platform, which increases client spend retention and generates incremental GSV and revenue.

Third, for our Upwork Standard offering, we charge clients a payment processing and administration fee, which was also introduced in June of 2016. This fee is generally 2.75% of client spend; however, clients can also opt to pay a small monthly subscription fee instead of the 2.75% fee if they choose to pay via ACH or automatic clearinghouse.

We have experienced an increase in ACH adoption over time since introducing this client fee. We view this as a positive trend for our business and our clients. ACH adoption lowers our take rate, but increases gross margins as we incur lower payment costs.

Our take rate was 14.3% in the third quarter as compared to 14.8% in the third quarter last year. This downward trend was expected, given what I just described. While we have many levers to increase take rate, we plan to be thoughtful about how and when we launch these features. We want to ensure that we are aligning our incentives with those of the freelancers and clients on our platform and that we are adding value when we introduce any new fees.

With these key operational metrics in mind, I will now turn to our financial results. Total revenue increased by 23% to $64.1 million. Marketplace revenue increased by 23% to $56.8 million and represented 89% of our total revenue.

Growth in marketplace revenue was driven by an increase in the number of core clients and higher client spend retention, which was evidenced by strength in our small business segment, growth in our U.S.-to-U.S. domestic marketplace offering, and an increase in direct sales from our enterprise offering.

One additional note is that in the third quarter, we started lapping the launch of our U.S.-to-U.S. domestic marketplace. And therefore, third quarter and our upcoming fourth quarter of 2018 have tougher comparables than our first-half results.

Managed services revenue increased by 21% to $7.3 million and grew at a slower rate year over year than our marketplace revenue. We expect this trend to continue in the coming quarters.

Non-GAAP gross profit increased by 23% to $43.7 million. Non-GAAP gross margin was 68%, remaining consistent with the third quarter of 2017. Gross margins are influenced by multiple components. First, the mix of revenue between our two offerings. Second, payment processing costs. This is our primary component of cost of revenue and increased by $1.3 million or 18% year over year, which grew slower than revenue.

Third, our spend on Amazon Web Services, which grew faster than revenue compared to a year ago as we are currently lapping the move of all of our services to AWS, which we completed in Q1 2018. Fourth, the cost of revenue for freelancer services to deliver managed services increased by 23% to $6.1 million, as more costly freelancer resources were used to provide managed services in this quarter.

In future periods, we expect cost of revenue to increase in absolute dollars, although the level and timing of revenue and cost of revenue items could fluctuate and therefore affect our cost of revenue and gross profit in the future.

Turning to operating expenses, sales and marketing expenses on a non-GAAP basis in Q3 increased to $18.6 million and represented 29% of total revenue compared to 26% last year. This increase was driven by investments to build out our enterprise sales team as well as marketing and advertising activities to drive brand awareness and attract new users. We intend to continue to invest in ROI-positive opportunities in sales and marketing to drive profitable growth.

R&D expenses on a non-GAAP basis in Q3 increased to $13.8 million and represented 21% of total revenue compared to 21% last year. Our R&D spend was focused on our mobile-first transformation as well as efforts to develop new products and features. We believe continued investments in R&D are important to further our strategic objectives.

G&A expenses on a non-GAAP basis in Q3 increased to $10.1 million and generated – sorry, represented 16% of total revenue compared to 14% last year. This increase was tied to our efforts to support our transition to a public company. We expect sales and marketing, R&D, and G&A expenses to increase in absolute dollars, although as a percentage of total revenue, it may fluctuate from period to period.

Our provision for transaction losses increased by $800,000 to $1.9 million in Q3 and consists primarily of losses from bad debt expense associated with our trade and client receivable balance and credit card chargebacks. Transaction losses increased due to the increase in trade and client receivables and related allowances, and we expect our reserves to increase as GSV grows.

Historically, transaction losses have fluctuated between 2% and 3% of total revenue. We focus on the trade-offs between increasing GSV and mitigating losses on our platform. And we are comfortable with the transaction losses in this range.

We incurred a net loss of $7.3 million for the third quarter compared to a net loss of $300,000 in the third quarter of 2017. Our basic and diluted net loss per share in the third quarter was negative $0.20 on 36.1 million weighted average common shares outstanding compared to negative $0.01 on 33.3 million weighted average common shares outstanding in the third quarter of 2017.

Our Q3 2018 net loss was largely driven by the remeasurement of our convertible preferred stock warrant liability of $3.3 million, which related to a warrant issued in 2013. The value of the convertible preferred stock warrant liability increased significantly as of September 30, 2018.

This was due to the proximity of our IPO and the final IPO offering price being significantly higher than the historical estimated fair value used to revalue the convertible preferred stock warrant liability. We have approximately $2.5 million of additional expense in Q4 2018 related to this warrant. Upon our IPO, our warrant liability was converted to additional paid-in capital.

We incurred a non-GAAP net loss of $1.4 million in Q3 2018 compared to a non-GAAP net income of $1.9 million in the third quarter of 2017. Our basic and diluted non-GAAP net loss per share in the third quarter was negative $0.04 compared to an earnings per share of $0.06 in the third quarter of 2017.

Adjusted EBITDA, a key metric for us in operating the business, was close to breakeven in the third quarter as compared to positive $2.8 million a year ago. We continue to balance investing in sustainable profitable growth while expanding our leadership position of this very large and expanding addressable market opportunity.

In April 2018, we established the Upwork Foundation Initiative to further our mission of creating economic opportunities to make peoples lives’ better. This program includes a donor-advised fund created through the Tides Foundation.

In May of 2018, the Company issued a warrant to purchase 500,000 shares of its common stock at an exercise price of $0.01 per share. This warrant is exercisable for 1/10 of the shares on each anniversary of the IPO. The proceeds from the sale of such shares will be donated in accordance with the Company’s directive.

There was no impact to our Q3 financials from this warrant. But based on today’s stock price of approximately $20 per share, we expect a non-cash charge recorded in G&A related to this warrant of approximately $1 million over the next 12 months, including approximately $250,000 in the fourth quarter of 2018. We plan to exclude the expense in deriving our adjusted EBITDA and non-GAAP net income on a go-forward basis.

Now to the balance sheet and cash flows. Our cash balances, funds held in escrow and escrow funds payable, trade and client receivables, and accrued expenses on our balance sheet and the related cash flow from operations are primarily impacted by the timing of funding on our escrow account.

Clients pay into an escrow account and payment is released to freelancers after services are completed, approved, and any relevant review period has lapsed. Escrow regulations require us to fund the escrow account with our Company’s operating cash if there is ever a shortage due to the timing of cash receipts from clients for completed hourly billings.

Freelancers submit their billings for hourly contracts to their clients on a weekly basis every Sunday. As of Sunday each week, we have not yet collected funds for hourly billings from clients, as these funds are in transit. Therefore, every Sunday we fund any shortage of cash into the escrow with our own operating cash and then collect this cash shortage from clients within the next several days.

Consequently, any quarter that ends on Sunday, like it did on September 30, 2018, and December 31, 2017, temporarily reduces our opening cash balances and cash flow from operations. To help fund the escrow account, we drew down $15 million at quarter end from a revolving line of credit, which we repaid on October 1.

Please note that the quarter ending March 31, 2019, and June 30, 2019, both end on a Sunday. And therefore, you should expect a similar impact on our balance sheet and cash flow from operations and for us to use the revolving line of credit in a similar fashion.

During Q3 2018, we used $18.6 million in operating activities, which was largely driven by the funding of the escrow with our operating cash on the last day of the quarter. We used $1 million in investing activities and had $15.3 million of cash provided by financing activities, mostly due to the draw on our line of credit of $15 million.

We recently signed a new lease for our Chicago office and are in negotiations for a new location in Silicon Valley. We expect to invest between $4 million and $5 million in capital improvements in the remainder of 2018 for these new office locations.

As a final point on the balance sheet and shares outstanding, we have provided a table in our press release that shows the pro forma basis impact of the IPO as if it had occurred as of September 30, 2018. We raised a net amount of $109.4 million in our IPO, which closed on October 5. We also repaid $10 million from our revolving line of credit in early October.

We had 36.9 million common shares outstanding at September 30, 2018. Upon the IPO, we issued 7.8 million shares of common stock and converted 61.3 million shares of preferred stock to common stock on a one-to-one basis. Therefore, as of the date of the IPO, we had approximately 106 million shares of common stock outstanding.

If we were a public company at the end of Q3, dividing our non-GAAP net loss of $1.4 million in Q3 2018 into the pro forma 106 million shares of common stock outstanding would result in a net loss per share of negative $0.01.

Turning to guidance. For the fourth quarter, we expect revenue in the range of $64.5 million to $66 million and adjusted EBITDA in the range of negative $750,000 to positive $250,000. We expect weighted average common shares outstanding to be in the range of 103 million to 104 million for the fourth quarter.

For the full year 2018, we expect revenue in the range of $250.5 million to $252 million and adjusted EBITDA in the range of negative $500,000 to positive $500,000. We expect weighted average common shares outstanding to be in the range of 52 million to 53 million for the full year 2018.

As an additional note, we are planning to provide revenue and adjusted EBITDA guidance ranges for the first quarter and the full year 2019 on our Q4 2018 quarterly earnings call. Thereafter, we will provide subsequent quarterly guidance and update our full-year guidance for revenue and adjusted EBITDA.

With that, I will turn it over to Stephane for some final thoughts.

Stephane Kasriel

Thank you, Brian. I am very excited about the trends we are seeing in online remote work. The world is now in the Fourth Industrial Revolution. This revolution is a chance to make a more positive future of work with more economic opportunity a reality.

Upwork is a leading force in this positive future of work. Our innovations bridge the skills gap to reduce friction in the labor market. At Upwork, we ourselves practice a work-without-limits model that includes a distributed team of on-site and remote employees. And we also engage with over 1,000 freelancers all over the world for our own specialized projects. It is amazing to have access to this highly skilled global workforce, which we believe to be a key component of our business success.

Again, thank you to our employees, the freelancers who provide services to us, the freelancers and clients who use our platform, our partners, and our investors. I believe Upwork can provide economic opportunities for millions of people. We are excited about the opportunities in front of us and we look forward to sharing this journey with you.

And with that, we are now happy to take your questions.

Question-and-Answer Session

Operator

Thank you. [Operator Instructions] Our first question comes from Mark May from Citi. Your line is now open.

Mark May

Thank you for taking my question and congratulations on your IPO milestone. I had a couple of questions. One on revenue retention. Thanks for some of the color around how you are thinking about the outlook there in aggregate.

But just wondering if there are any interesting differences in revenue retention growth rates by different types of clients, if it’s employers with fewer than 100 employees or greater than 100 employees. Or any other ways that maybe you slice and dice the numbers, if you see any sort of interesting differences and different types of client types.

And then you’ve got a number of areas of investment and growth opportunity; enterprise comes to mind, brand, marketing comes to mind. How are you thinking about weighing investing in these growth opportunities versus trying to manage kind of near-term profitability improvements? Thanks.

Stephane Kasriel

Sure, great question. Thank you. So I would say the answer to the first question, if I had to like say it in one line: the bigger the client is, the more they retain. So the way we look at it is we have four segments of customers. We currently segment clients based on number of employees.

So we have what we call very small businesses, or VSBs, which are companies with less than 10 employees. Then what we call small business, which are companies between 10 and 100; in-market, between 100 and 1,000 employees; and large businesses above 1,000. And for each segment, the retention rate goes up.

One of the things we talked about in the S-1 was how we look at core clients versus noncore clients and the difference between logo retention or actual retention of individual clients versus dollar attention. And if you remember what we said back then, essentially the overall logo retention – and this is 2017 data, but the overall logo retention for the Company was 58%. The logo retention for core clients was 83%.

The dollar attention overall, and it’s not very different for core clients versus noncore clients, but the dollar attention overall is 106%. So what happens is 80%, roughly 80% of our business is with these core clients who tend to be bigger companies.

To give you an example, when you sign up with a gmail.com email address, if you don’t have enough money to buy a custom domain name, you probably won’t have a ton of money to spend on Upwork. So those tend to be one-and-done; they don’t tend to spend a lot of money.

The ones that end up spending $5,000 and above, which are the ones that we call core clients, they still churn because small businesses go out of business. But the ones that remain end up growing over time. We grow with them. As they are more successful, they spend more money on the platform.

So that 83% core clients logo retention, the way it translates into 106% dollar attention is because the ones that do remain – so the 83% that do not churn – spend 106% divided by 83%, which is about 125% year on year.

The bigger clients, like if you look at the enterprise business, they tend to grow on their own and they grow even more when our sales team gets involved. So at this stage, we do not have enough salespeople to handle all of the leads that are coming in to the system, which is why we continue to hire aggressively in sales. We see that when salespeople handle an account, the account is more likely to grow more substantially.

To answer your second question, the question around how we think about investments, I would say – I would separate sales and marketing from product and engineering. So in sales and marketing, we really look at marginal CAC to LTV ratios, if you will.

So if you look at marketing as an example, we spend money on online marketing channels like Google, Facebook, LinkedIn, Microsoft, etc., etc. When we buy an additional keyword or we launch an additional campaign, we are able to measure what the cost of acquisition looks like. We look at early trends in terms of spend and project data to look at what LTV looks like. And we continue to invest up to the point where the marginal ROI or the marginal NPV, if you will, is below a certain threshold.

In sales, it’s kind of the same thing. We hire new salespeople. We give them some time to ramp up. And as long as the new cohort of salespeople we have added keeps their numbers at a rate that we feel comfortable with, then that means the productivity is good, the ROI of the investment is good, and we continue to hire more people in the new cohort. If at some point we realize that the new cohort is not making their numbers, then we would slow down on hiring and go figure out what is broken and go address it.

In R&D, it’s a little bit different. It’s a portfolio approach. We look at some investments as being quick wins. If we can fine-tune an algorithm to improve time to hire or improve the clearing rate of the transactions or what have you, we will look at those on a pretty tactical basis. We do a lot of AB testing; we have a lot of capabilities to test various user interfaces, treatments, various algorithms.

But then we also make bigger bets. When we are looking at – when we were looking, as an example, at launching our U.S. domestic website, there was some significant community risk associated with it. There was some significant business risk associated with it. And we take the things with, you know, I would say a portfolio approach, the way an investor would look at it as a portfolio approach.

Mark May

Perfect.

Stephane Kasriel

Great question.

Mark May

Thanks a lot.

Operator

Thank you. Our next question is from Mark Mahaney from RBC Capital Markets. Your line is now open.

Mark Mahaney

Thanks. I wanted to ask about the domestic-to-domestic or the domestic marketplace and how that has rolled out for you in the U.S. I think you are also rolling that out in the UK. Just talk about the materiality of that, learnings you’ve had from that, and your optimism in terms of being able to roll that out to multiple international markets. Thank you.

Stephane Kasriel

Sure. So there’s kind of four things we have been doing and are continuing to do in our domestic initiative. So the first one, which we launched last year, was the U.S. domestic website itself.

And as a reminder to explain what it means is when you sign up on Upwork today, either as a client or as a freelancer, if you state that you are based in the U.S., we give you a choice between participating in the overall global upwork.com marketplace or participating in the U.S.-only marketplace.

And what that means is as a U.S.-based freelancer, if you choose the U.S.-based marketplace, you will only see the subset of jobs where everybody else who is bidding on the work is based in the U.S. And as a result, you are not competing against people overseas, people who potentially have a lower cost of living and therefore might be bidding at rates that are lower than something that you are comfortable with. On the client side, what it means is when you choose the U.S. marketplace, you will only see the subset of freelancers who are based in the U.S.

Now, as you can imagine, U.S.-based freelancers love the feature. The idea that you – if you are based in San Francisco, you are competing against people in New York and Chicago, but you are no longer competing against people in Greece or Chile or Russia makes a ton of sense for you.

So we have seen tremendous amounts of adoption from the freelancers because they are bidding against other freelancers in the U.S. They have been able to increase their rates substantially. So we are seeing the hourly rates on the U.S. marketplace be substantially higher than the global marketplace. And they have gone up over time as people learn what the new rules of the game look like, if you will.

On the client side, you could say, well, why would a client choose to limit their options? Why would they want to hire only in the U.S. if they could hire globally? And there is any number of reasons, but I would say, back to Mark’s question earlier, the bigger companies tend to be a little bit more risk-averse, a little bit more conservative than the very small companies. And so what we see is as our business grows faster with bigger companies, we see more and more interest in this type of domestic hiring.

So it is one of those things that’s been a huge win-win-win for the Company, if you will. The freelancers in the U.S. are happy. They are getting jobs at a higher rate than they did before. We are signing up incremental clients that were not interested in the Upwork website before. They are posting better jobs that pay higher. And of course, because we are a take-rate business, we end up making more money.

So some of the U.S. domestic website is incremental; some of it is, quote, unquote, cannibalistic, meaning these are jobs that potentially would have been posted on the global marketplace and instead they get posted on the U.S. marketplace. When that happens, though, because the hourly rate is higher, it still ends up growing GSV.

So that is what we did last year. We continue to optimize it; like we do all sorts of things to make it continue to grow faster. It is going substantially faster than the global marketplace. As we have mentioned before, the U.S. is both our number one geography for clients and also the number one geography for freelancers and the number one corridor.

So if you look at pairs of countries between clients and freelancers, the U.S. is also number one for that and it is growing substantially faster than the overall business. So that’s one thing we have done and continue to invest in.

A second one was going after metropolitan areas about hiring. So once we realized that we had enough network density to provide liquidity at the U.S. level, the next logical question was can we do this for the San Francisco Bay Area, can we do this for New York and the local New York City area, which are two of the biggest metropolitan areas for us in the U.S.

So we launched that. It is early, I would say. It is getting really good results. And I would say one of the interesting things about it is that even if clients state that they would prefer somebody in the Bay Area, sometimes we don’t have the supply that they need, but we end up finding them somebody else in the U.S. They end up hiring anyway, so the job gets filled and we have increased the top of the funnel. So it seems to be working pretty well, even though obviously as you get to more and more local hiring, finding the perfect match between supply and demand is that much harder.

There is a third path, which we haven’t started yet, but I’m kind of like giving you a sense for where that is going, is even more local hiring. So you imagine that for some jobs, you need the person to come on-site if not every single day, at least part of the time.

So to give you an example, one set of categories that we have been doing historically is video editing. When you do video editing, whether it is adding voice, special effects, video production, etc., etc., a lot of that work can be done remotely. However, at some point, you need a camera crew to come on-site and shoot the video.

And that historically was not something that was easy to do on Upwork because we tended to really focus on remote type of work. So we are going to do a soft launch, a beta if you will, in certain categories in certain markets to see whether we have enough supply, enough demand, and enough interest by our customers to test an even more local type of hiring.

And if you think about it, the way I would look at this is all of these are concentric circles where the market size is increasingly larger, but our strength in that market is potentially increasingly lower. And therefore, we would expect share of wallet to be lower. So it’s a smaller percentage of a bigger market versus a bigger percentage of a smaller market. And ultimately, ideally, we want to own all of that.

The fourth thing, which I think is your question, Mark, is around the UK. So we launched a what we call domestic-light – and I would say I would really underline light here – in the UK, where the idea was we took the functionality we built for the U.S. and essentially launched it more or less as-is for the UK.

The UK is one of our top markets, both on the buyer side and on the freelancer side. As is a little bit obvious, it’s an English-speaking market, which makes localization much easier than if we wanted to do a test in, let’s say, Japan, France, or Germany.

And the reason why I would call it very much a light approach is because obviously our system today is in U.S. dollars and so we do not support the British pound. And so the awkwardness, I would say, of that pilot is that we are enabling a business in the UK to engage with a freelancer in the UK to pay each other in U.S. dollars, which is not ideal.

So we are hoping to get some early read in terms of what it does to client demand, what it does to freelancer supply, how happy people are with it. We fully expect to get a lot of feedback on what is working and, more importantly, what is not working in that solution. At the very least, the obvious one is if we want to do this at scale and do this really seriously, we need to start supporting multiple currencies, which is a significant investment from a product and engineering effort.

As you may remember, we are regulated here in the U.S. as what is known as a licensed escrow agent. I am not totally sure that the U.S. regulator really wants to deal with multiple currencies. This country tends to think that the U.S. dollar is the currency for the country. And so having a multicurrency escrow account might be challenging.

Mark Mahaney

Okay. Thank you very much, Stephane.

Operator

Thank you. Our next question is from Brent Thill from Jefferies. Your line is now open.

Brent Thill

Good afternoon. Stephane, just on the enterprise business, I was just curious if you could give us some more mile markers of where you are at, kind of how you see that business unveiling and rolling out over the next year. You mentioned you don’t have enough salespeople. Are you going to step on the gas there in terms of hires here going forward? And then I had a quick question for Brian as a follow-up.

Stephane Kasriel

Sure. I mean, I think the big long-term picture here is if you think about the underlying theme and the types of, I would say, incumbent providers that this Company is disrupting, they tend to do most of their business with large companies.

So we have started, I would say, in classic disruptor playbook. You start with the small companies that nobody else is able to serve and then progressively you go after where a big chunk of the money is and you go after the bigger companies.

So in the very long term, if you look several years out, the majority of this business will be in the enterprise space. But it’s starting from a small base. Less than 20% of our business is done with companies with more than 100 employees. So even though it is growing faster than the overall business, it’s going to be several years before it becomes 50%-plus of the business.

What’s happening today is we have a relatively small sales team and people are hitting their numbers. We are happy with the productivity levels, even though we will continue to put pressure on people and increase quotas and make people even more productive over time. But generally, it is working. The payback periods and the ROI is good and so we are continuing to invest.

To give you an example of this, we just signed a lease in Chicago for an office that can contain up to 500 people. Today we have about 100 people in Chicago, so we are definitely planning to expand quite a bit.

At the same time, just to make sure I set proper expectations here, you don’t just double a sales team overnight. This is truly – you need to build the management bench strengths, you need to manage the processes to be able to do this at scale. So we will continue to hire as fast as we think we can hire.

I don’t think at this stage that we are limited by the number of leads. We are limited by the ability to hire great talents and onboard them and allow them to be successful, including having all of the facilities or, if you will, the different stakeholders that help a salesperson be successful at Upwork.

Brent Thill

Great, thank you for that. And Brian, I just wanted to clarify. In the first half of the year, you had high-20% growth. You dipped into the lower 20%s and you are guiding now back into 25% growth I think at the midpoint for Q3.

So I am just curious. Was there any seasonality or anything unusual in Q3 that brought the growth down? Or are we just trying to parse the numbers too much and the overall trajectory looks pretty healthy from your perspective?

Brian Kinion

Yes. Stephane mentioned about the U.S. domestic-domestic site that was launched in Q2 of last year and it’s mostly lapping related to that. So we always try to give some color on that from that perspective.

And as we have talked in the past, like, Q2 to Q3 are usually about similar in that there is seasonality in those business and you tend to see a little bit of acceleration in the fourth quarter over Q3. But you have that seasonal sort of billing that we’ve talked about in the past.

Brent Thill

Okay. And nothing in terms of unusual on the take rate? I know it dipped down to one of the lower levels you’ve seen. Do you expect a stabilization around the low 14%s? Is that a reasonable level to think about in the models going forward?

Brian Kinion

It moves around a lot and that is why we have tried to model – provide guidance on the revenue number and not on GSV and take rate. The mix of business will move around, as we indicated; like, the managed services business probably will grow slower than the marketplace.

We are seeing more and more in the 5% tier, which we see actually as a positive, as well as more and more ACH adoption. Both of those are actually benefiting us from the perspective that we are seeing less circumvention than we saw before the pricing change. But those three elements will move around a little bit and so there will be a little bit of downward pressure.

Stephane Kasriel

The thing I would add to this is we ourselves measure ourselves mostly on revenue, to be honest. The thing that I think matters a lot, like, if I were looking at this as an investor in the business, what I would worry about more is if take rates start to go up tremendously. This would be a case where you are saying, hey, this company is struggling to create a lot of value for its users and it’s overmonetizing as a way to compensate for it to try to jackup revenue numbers.

We are in the reverse situation right now, where we are letting, on purpose, take rates decline. We are not planning to make major changes to the pricing because we think that the incremental GSV that we get at, even at a lower take rate, still generates more revenue, generates a lot of satisfaction from the customers. And that is what is driving this spend retention rate going up, the number of core clients going up.

Generally, this is a very healthy business where people are spending more and more, they are retaining more, they are referring more people to us. And we are pretty happy with the evolution of revenue versus take rate right now.

Brent Thill

Great. Thank you.

Operator

Thank you. Our question is from Ron Josey from JMP Securities. Your line is now open.

Ron Josey

Great, thanks for taking the question. So two really quickly and the first one is just on escrow. And Brian, I know you talked about that earlier. But specifically, I wanted to ask about the competitive advantage that you believe you have on escrow.

Only because in talking to a bunch of freelancers, everyone sort of highlighted the benefits of the escrow process that Upwork has. And so can you just help remind us why – how difficult it might be to replicate your escrow process, particularly as you process payments across so many countries? So that is one.

And then the second question is just on advertising. And if you could talk about brand awareness, the success in your ad campaign thus far, and maybe how that ties to the adoption in the domestic marketplace. Thank you.

Stephane Kasriel

Sure. So I would say escrow, first of all, is really hard to do, and secondly, it is a huge competitive advantage. So I would say competitive advantage compared to what? So first of all, compared to traditional freelancing.

So most freelancers at this stage do not work through an online platform like Upwork. We estimate in this study that I mentioned, Freelancing in America, we estimate that freelancers in the U.S. earned about $1.5 trillion last year. So obviously, online freelancing through Upwork and other platforms is only a small percentage of total earnings.

If you ask traditional freelancers like what is their number one top-of-mind like problem, they would say getting paid. The number of freelancers who don’t get paid, get paid less than they expected, get paid late is huge.

If you think about companies, companies treat freelancers like they treat their other vendors. The game of trying to push working capital off as far away in the future as possible, paying your freelancers like vendors, not like employees, is a huge issue. So compared to traditional off-line freelancing, this ability to know that you always get paid on time and you will always get paid the right amount is a huge part of the value prop.

There is a separate question there, which is competitive advantage compared to other companies that might do what we do. And I would say the fact that we voluntarily decided to get regulated by the Department of Business Oversight, it’s a huge hurdle, it’s painful, it’s not easy to deal with having two sets of books and two annual audits and all of that stuff. It requires a level of precision in your numbers that is down to the penny, which, when you do $1.5 billion per year, is actually a pretty high level of precision.

But once you have it, you can market it heavily. It’s very visible on our website. We are the only – to my knowledge, we are the only significant player in the online freelance place that is regulated by the DBO. In fact, there is a company in our space, publicly traded in Australia, that, amongst many others, just got a cease-and-desist by the state of California related to the fact that they are not regulated like we are. And they need to stop operating their service in California.

And then your second question was on brand awareness. And I would say like – so we definitely measure aided and unaided brand awareness. We measure before and after. We measure in comparable cities ones where we did the campaigns, ones where we don’t do the campaigns. And we are definitely seeing early trends that it has an impact on awareness and consideration.

I think just to set the right expectations here: the amount of money we spend on this right now are very small compared to some of the other players, both in our industry and in others. Think about Indeed is spending or what LinkedIn is spending, or in slightly different industries what a Go Daddy, as an example, would be spending. We are orders of magnitude lower than everybody else and it’s also relatively recent for us.

And we have literally started doing this about a year ago. It takes a long time to get to a level where you become a recognized brand name and people think of you as a top-of-mind place to post their jobs.

So this is going to be an ongoing and a long-term effort for us. But I would say at this stage we are very encouraged by the fact that it A, seems to be working for everybody else. And because we are serving similar customers, we expect that it would also work for us. But B, early signs seem to indicate that it is actually doing pretty well for us as well.

Ron Josey

Great. Thank you.

Operator

Thank you. Our next question is from Scott Devitt from Stifel. Your line is now open.

Scott Devitt

Hi. Thanks for taking my question. Was just wondering with the state of the U.S. job market approaching full employment and more jobs being available than workers to fill them, how that may play into the corporate adoption curve for your services? Or just more generally how it may impact the business. Thanks.

Stephane Kasriel

Sure. I think our business – there’s two parts of our value prop that we will emphasize differently depending on where we are in the economic cycle. So in today’s economy, even though we provide cost savings compared to traditional ways of doing work, that is not the number one reason why most companies use us.

People use us for access to talent. They use us for the ability to hire people much faster. A typical job in the U.S. right now stays open for 30 days. The median time to fill a job on Upwork today is less than a day. So there is that part of the value proposition is really important.

I would say one thing that people tend to underestimate is there are a lot of people that are college educated, highly skilled, and whose skills and time is underutilized right now. And to give you just one data point: in that survey we just completed, Freelancing in America, one of the questions we have been asking in that survey every year is how much money would a traditional employer have to pay you in order to take a full-time job.

And literally every year, the answer is 50% of people say no amount of money would get them to take a full-time job. These are people that are not accessible to you as an employer if you are just looking for full-time employees.

This year, we went a little bit further and we were trying to understand like – this sounds crazy, right? How could 50% of survey respondents say that no amount of money could convince them to take a full-time job? And that just seems to defy common intuition here.

So we asked additional questions to try to understand, like, what is driving this. Why would somebody really not consider a full-time job? And what we heard back is about 42% of respondents are saying they have things in their personal lives that prevent them from taking a full-time job.

And typically, this was around physical and mental health. It was around care, either young children or elderly parents or a sick spouse, and geography. People live in a place in the country where there are no good economic opportunities and for one reason or another, they are unwilling or unable to move. And so that’s actually 42% of the freelance population that is saying, look, I am not available for a full-time job, but I am fully available for freelance job.

So part of the answer to this full occupation economy and local skills gap is you need to open up new channels. You as an employer need to look at different ways of hiring people. And I think we can be a very powerful way of expanding the size of the labor force in an economy like this, where there are fewer traditional jobseekers than there are open full-time positions.

Scott Devitt

Thank you.

Operator

Thank you. Our next question is from Nandan Amladi from Guggenheim Partners. Your line is now open.

Nandan Amladi

Hi. Good afternoon. Thanks for taking my question. So as you approach building out your network with the different types of skills that you have in the system, how do you decide whether to go broad in terms of expanding the types of skills versus deep, where you have like levels of expertise? And how do you decide what skills to add?

Stephane Kasriel

Sure. So maybe to start with your second question, I mean, we don’t really decide. In a way, it is our customers that decide for us. So ultimately there is some engineering work to enable new skills and new categories, but typically either people explicitly tell us, like, I have this skill and I think you should be adding it to the platform. Or I am looking for this skill and you should add it to the platform.

Or the machine learning algorithms will pick it up themselves. Like literally, we will see keywords that seem to be skills that we don’t know about. And that gets brought up to a category manager within our Company who looks up at what is going on. And if that seems to be a meaningful trend and something that we think we should be supporting in the Company, then we add it to the product and make it available to the entire customer base.

Now to answer the first part of your question, historically, we haven’t gone very deep. Historically, a skill on Upwork or a category on Upwork, if you will, is a label. So when you post a job, you can say this is a web development job and I require PHP as a programming language and I require MySQL as the database. And the machine did not particularly understand what MySQL meant or what PHP meant.

Our categories initiative, which is a big multi-quarter product development for us, is around deepening and verticalizing the platform. I mean, fundamentally, the way you hire designers or the way designers promote themselves is radically different from the way you hire a strategic consultant or the way a strategic consultant markets themselves.

So you are going to see us progressively have taxonomies that are truly structured where we really understand what these things mean. And from a user experience standpoint, you will see the experience for hiring for different types of roles be different as we verticalize parts of the site.

Now, we can’t do this overnight across 70 categories and 5,000 skills. So you will see us do it in places where we think it is going to have the most impact, both because of the size of the category as it is today and the increase in value proposition that we think verticalization would bring.

Nandan Amladi

Thank you.

Operator

Thank you. Ladies and gentlemen, thank you for participating in today’s conference. This concludes today’s program. You may all disconnect. Everyone have a great day.